A triple GEM detector for LHCb

99-024 The full size (23 cm $\times$ 25 cm) prototype of a triple GEM detector worked well in a hadronic beam. Extreme high gains ($>$ 20'000) with almost no sparking could be reached. But it is possible that a single GEM spark can lead to a short in a GEM foil and destroy the detector. It seems, that a carefull conditioning of the detector during strong irradiation, results in a much more stable operation. Fast signals from Fe$^{55}$ and MIP have been recorded, in the latter the clustering of the primary ionisation charge has been observed. The influence of the shaping time of a simple gaussian filter on the detector signals was studied. If the pulse time is defined by the pulse maximum a longer shaping time improves the time resolution. A longer shaping reduces the observed cluster width as well. The optimal electronic shaping time seems to be of order 15 ns for such a simple filter algorithm.